Forecast / Logistics

Access from: Forecast → Logistics

AI-based stock forecasting tool for planning Amazon FBA shipments.

What does Epinium generate?

Epinium does not automatically create forecasts or replenishment orders in Amazon. The platform calculates expected demand and generates an Excel file with the recommendations:

• Sellers: use the Excel as the basis to prepare the upload file required by Amazon Seller Central to create the inventory forecast (e.g. Supply Chain by Amazon).
• Vendors: use the Excel as an internal production control and supply planning tool.

Create a forecast

• Select Amazon connections (which accounts to include)
• Define minimum final stock (safety threshold)
• Define minimum units per shipment
• Regional configuration (different marketplaces)
• Include FBM stock: Optional parameter to also consider sales from the FBM versions of the same FBA product. When enabled, the forecast model accounts for demand generated by the FBM channel of the same product, obtaining a more accurate estimate of total demand

Forecast results

Table by product and marketplace with:

Field	Description
Current FBA stock	Inventory available in Amazon
Forecasted demand	Next 30 / 60 days
Recommended units	How many units to ship
Suggested dates	When to make the shipment

Export

Download the forecast as an Excel file. The file is not uploaded directly to Amazon:

• Sellers: use it as the basis to prepare the upload file and create the forecast in Seller Central.
• Vendors: use it as an internal production control and supply planning tool.

How the forecast works

The model combines two historical data sources to estimate expected demand:

• Recent sales: sales from the same number of days as the forecast period, counted backwards from the start date. If the forecast spans 30 days, the model uses the previous 30 days as a reference for the current trend.
• Seasonal sales: sales for the same date range (same day and month) from the previous year, as a reference for seasonal behavior.

From both sources, the model automatically assigns a higher weight to whichever predicted the most recent period more accurately. If last year's seasonality was more representative than the current trend (or vice versa), the calculation reflects that without any manual configuration needed.

Why might a forecast not be generated?

If either reference window lacks sufficient data — for example, a product launched less than a year ago or with gaps in sales history — the forecast will not be available for that product. To improve coverage, make sure products have sales history both in recent months and in the same period of the previous year.

Workflow: plan stock replenishment

1. Go to Forecast → Logistics
2. Select connections and configure parameters (minimum stock, minimum units)
3. Generate forecast
4. Review the recommendations table by product
5. Export to Excel
6. Sellers: use the Excel to prepare the upload file and create the forecast in Seller Central Vendors: use the Excel for internal production control